Land surface conditions affect radiation processes, atmospheric temperature and precipitation through heat and moisture exchanges between land surface and the lower atmosphere. Therefore, in order to evaluate the heat and water budgets and the water cycle properly, a sophisticated land surface model is important for the NWP model.

The JMA’s operational global NWP model (JMA-GSM) currently adopts a Simple Biosphere (SiB) model. The operational SiB model has a systematic error of an overestimation of thaw. Moreover, it treats the soil processes so simply as not to simulate a formation of frozen soil properly.

JMA is developing a new SiB model, which is more sophisticated than the operational one in terms of snow as well as soil water and temperature. The long-range forecast experiments with the operational and the new SiB models show that the new SiB model predicts a snow cover area reasonably in a melting season while the operational one predicts a snow melting excessively. However, the new SiB model slightly overestimates a snow cover area over Russia and North America in the early winter.

In order to increase comprehension of land surface phenomena and the global water and energy cycle, CEOP (Coordinated Enhanced Observing Period) was launched in 2001, which is striving to achieve an integrated database of common measurements from both in situ and satellite remote sensing measurements, as well as model output. A part of the in situ data for EOP-3 (from October 2002 to September 2003) is released from the CEOP data archive center at University Corporation for Atmospheric Research (UCAR). We carry out the assessment of the model forecasts with each SiB model against the CEOP data and reveal that the incoming long wave radiation is underestimated in both models.

JMA also develops a new radiation scheme, which represents the absorption by water vapor in the troposphere better than the operational one. The comparison between the radiation schemes against the CEOP EOP-3 data shows that the new radiation scheme significantly reduces the underestimation of incoming long wave radiation. A long-range forecast performed with both the new SiB model and the new radiation scheme shows encouraging results such that the overestimation of the snow cover area in the early winter is reduced and that the thaw is adequately predicted in spring.

The diurnal cycle of the model forecasts is examined and the results are also reported.